"Pirate" electrolytic caps - identifying?

I'm trying to analyze a group of almost simultaneous failures in
products that were released in 2002. The product is built on an x86
motherboard, in a housing that meets or exceeds normal PC-case
airflow. It SEEMS - and I'm still burning in a "repaired" board to
test this theory - that the culprit parts are specific electrolytic
caps used in various places over the PCB.

The caps in question are G-LUXON 1000uF, 6.3V, 105 deg C rating, date
code 0117(M). ALL of them are visibly bulged on top. Sticky dust-mud
has gathered in the intersection of the score-marks on top, which
leads me to believe that they bulged enough to break the can here and
electrolyte has vented out. All the other electros on the board look
normal. But then none of the other caps on the board are G-LUXON
brand; they are mostly Sanyo.

I'm wondering if this is a symptom of that problem people were talking
about a while ago, where an incomplete/not-ready-for-primetime
electrolyte formula was stolen and used in Chinese-made caps, which
then exhibited an unusually high mortality rate.

 

Almost certainly. Luxon was reportedly one of the companies
(Taiwanese) which purchased the bargain-priced (but defective)
electrolyte from Luminous Electric in China.

I had a pair of Apple Airport base stations go bad as the result of
failed Luxon 'lytics. Neither had yet leaked, but both were bulging
badly and had gone high-ESR. Replacement with another brand fixed the
problem.

You can probably expect most of the products built with the defective
capacitors to fail, after a few hundred to a few thousand hours of
operation, as the caps electrolyze away their electrolyte. As far as
I know, replacement of the defective caps is the only preventive or
curative measure.

 

FWIW, Luxon (Taiwan based, but undoubtedly with China manufacturing)
has denied involvment in that scandal.

http://group.luxon.com.tw/



Best regards,
Spehro Pefhany

 

Seems strange that they all went south, is it not also possible that the board
had an on-board regulator failure??

Just a thought.

Peter

 

And Clinton didnt inhale. Pale of broken caps from Abit boards knows
better.


Pozdrawiam.

 

The caps in question are G-LUXON 1000uF, 6.3V, 105 deg C rating, date

It doesn't seem very likely. These caps smooth a 3.3V rail which runs
all over the board. There is a group of four of these caps near the
processor, and one near the RAM slots. Near a couple of other
subassemblies, there are smaller (330uF 6.3V) caps on the same rail.
These other caps are visually fine, and I pulled and tested a few at
random and found them close to nominal value. It's just the G-LUXON
1000uF caps that are bad. Most of them measure 5-10uF or open circuit.

 

The caps in question are G-LUXON 1000uF, 6.3V, 105 deg C rating, date

The runtime clocks on the failed units are, interestingly enough,
clustered around an average of approx. 3,400 hours. Without looking
through them all, I see the lowest at hand is 2171 and the highest is
3709.

Well, this is good and bad news, assuming that is in fact the only
problem. It's good news in that the customer obstinately selected these
particular boards instead of a more expensive board I had originally
recommended . It's bad news in that I think I'll be asked to repair
them all rather than testing and requalifying some new motherboard.

One of those "quick" contracts that rears up a couple of years later to
bite one's hindquarters. Oh well. Thanks for the reply. I hope the one I
repaired is still running on Wednesday, because if so I think the
problem's fixed (it usually shows up within an hour of powerup after
cold, I figure a few days trouble-free is fairly safe).

 

On Sun, 31 Oct 2004 18:50:57 +0000 (UTC), the renowned Peter A Forbes

If one goes high ESR, it increases the power dissipation in the
remaining good caps.


Best regards,
Spehro Pefhany

 

It's because the defective electrolyte produced gas at a consistent
rate.

Of course, it's likely that a regulator failure would fry the board
BECAUSE the capacitors failed.

 

They are usually used as the filters on a switching power supply.

 

: > >The caps in question are G-LUXON 1000uF, 6.3V, 105 deg C rating, date
: > >code 0117(M). ALL of them are visibly bulged on top. Sticky dust-mud
: >
: > Seems strange that they all went south, is it not also possible that the board
: > had an on-board regulator failure??

: It doesn't seem very likely. These caps smooth a 3.3V rail which runs
: all over the board. There is a group of four of these caps near the
: processor, and one near the RAM slots. Near a couple of other
: subassemblies, there are smaller (330uF 6.3V) caps on the same rail.
: These other caps are visually fine, and I pulled and tested a few at
: random and found them close to nominal value. It's just the G-LUXON
: 1000uF caps that are bad. Most of them measure 5-10uF or open circuit.

Are you sure you didn't touch the ripple current specs of the caps? I guess,
near the processor and the switching regulator the caps will carry a
substantial ripple current.

 

For your pleasure and enjoyment, an excerpt from an old comp.risks
digest:

--- Begin Included File ---
------------------------------

Date: Tue, 20 May 2003 16:44:19 -0400
From: "Jay R. Ashworth" <>
Subject: The Great Capacitor Scare of 2003

In RISKS-19.13, Mich Kabay quoted the *EE Times* on "The Great
Capacitor Scare Of 1997". People were building motherboards without
enough power supply filter caps, it seems, and machines were locking
up.

Oh, to have problems that minor again...

The Great Capacitor Scare of 2003 is going to be *much* worse.

It seems, according to several news stories (linked at the end) that a
materials chemist who worked for a Japanese company, Rubycon
Corporation -- which manufactured electrolyte for electrolytic (!
capacitors -- left his employ, and ended up working for a Chinese
capacitor maker, Luminous Town Electric. (These names tend to sound
quaintly amusing to USAdian ears, which might not be accidental...)
Apparently, in a fairly clear case of corporate espionage, the
fellow's cow-orkers then "defected with the formula" (PCN says, in a
confusing bit; defected to where he was?), and began to sell the
electrolyte to many Taiwanese capacitor makers.

Alas, there was one small problem.

The formula wasn't *complete*. The capacitors, which ought to have
been good (in some cases) for up to 4000 hours, were failing in half
that -- or, if you believe Intel, in as little as 250 hours. The
electrolyte apparently outgasses hydrogen, and pops the seals on the
cap, leaking electrolyte onto the board. The missing ingredient was
the one which prevented this. I'd speculate that this might not be a
point-catastrophic failure... these caps might pop and leak out
slowly, shorting out circuits.

But it's even worse.

The Inquirer may put it best:

"It is not currently known how many market segments may have been
affected by these poor parts, which can be found in motherboards,
switchmode power supplies, modems and other PC boards. The
failures of the aluminum capacitors might just be the 'tip of the
iceberg,' says Zogbi. "Other component failures from low-cost Asian
suppliers might be forthcoming," he warns.

"Around 30 per cent of the world's supply of aluminum capacitors is
manufactured in Taiwan, according to the Paumanok Group. Confusion
over which manufacturers may have used the faulty electrolyte is
sending buyers back to Japan to source their capacitors. The
extent of the problem in product that has already shipped won't become
clear until components start failing, which may not happen until
halfway through the products' life expectancy. "

But even *that* may understate the problem...

How many electronic products do *you* know of that use electrolytic
capacitors? The RISKS are so obvious that I don't even have to say
"The RISKS are obvious". [But you did anyway! PGN]

*The Inquirer* coverage is at http://www.theinquirer.net/?article=6085


*Passive Component News* is at http://www.niccomp.com/taiwanlowesr.htm

Check out the tenor of the editorial footnote; it's as classic as it
is uncommon.

TTI, who bill themselves as "The world's leading distributor of
Passive, Interconnect, and Electromechanical components" have put up
an entire page tracking press coverage of the issue:
http://www.ttiinc.com/MarketEye/Aluminum_Cap_Issue.asp

Jay R. Ashworth, The Suncoast Freenet, Tampa Bay, Florida
http://baylink.pitas.com +1 727 647 1274

--- End Included File ---

I haven't checked to see if the referenced links are still up. FWIW,
yours is the first problem regarding these electrolytics (if your
problem _is_ these electrolytics) that I've heard about.

Regards,

-=Dave

 

My thoughts also. If this is indeed a 3.3V line for a power-hungry processor
the caps needs to be very low ESR types or many of them in parallel

Cheers

Klaus

 

Usually *both*.


Best regards,
Spehro Pefhany

 

: It doesn't seem very likely. These caps smooth a 3.3V rail which runs

I didn't design the board - it's an off-the-shelf Pentium III
motherboard. These systems are due for replacement in a little less
than nine months. I figure the new caps ought to last this long, since
they previously lasted more than two years.

 

I first read about the cap problem in IEEE Spectrum. The online version
of the article is here
http://www.spectrum.ieee.org/WEBONLY/resource/feb03/ncap.html

That article specifically mentions Luxon caps, but also says that the
company denies producing any faulty caps.

Ed

 

In the name of completeness, at least two other explanations are possible:
1. They really didn't produce any bad parts and the reports are wrong.
2. They did produce bad parts but were/are not aware of it.

However, personally, I find your expanations more plausible...

Ed

 

They could be lying or they could have been counterfeit components (a
definite risk in China).



Best regards,
Spehro Pefhany

 

Flawlessly logical, Mr. Spock, err, Ed.

But I was taking into account the many reports on the web of Luxon
branded parts failing (at least one of which is from someone I know
not to be an agent of their competition. Also, I believe they actually
denied using electrolyte made from the stolen formula. You may still
be able to find the actual statement from the company (it used to
exist on their web site) archived on the net (I did when I looked). If
they actually denied producing any faulty parts at all, ever, you'd
*know* they were lying. ;-)


Best regards,
Spehro Pefhany